UrbanAI: Urban Energy Solutions Using AI

Cities have an outsized environmental impact. They are a significant contributor to climate change, accounting for an estimated 75% of global CO2 releases. Cities also consume 78% of the world's energy and house more than half of the worldĺs population. All this occurs in an area that is less than 2% of the Earth's surface. With ongoing urban growth and industrialization, these numbers will grow. One of todayĺs significant challenges is understanding how we can reduce the energy and environmental impacts of our cities. This issue has many aspects, including protecting biodiversity, reducing greenhouse gas emissions, and transitioning to green energy sources for building heating and cooling. Other problems include adapting urban infrastructure to climate change and promoting energy and environmental justice, especially in underrepresented and underprivileged communities. Developing technologies that allow us to measure, analyze, and predict urban processes related to social dynamics, energy use, and environmental impacts is vital to addressing these problems. We will develop a cloud-computing AI/ML (artificial-intelligence/machine-learning) software called UrbanAI for near-real-time data streaming, analysis, interpretation, and mapping of urban socioeconomic, demographic, health, energy, environmental, and infrastructure data. UrbanAI will focus on integrating disparate data, predicting energy needs and utilization with AI/ML, and suggesting applications to expand the use of green energy technologies. We will utilize transfer learning to propagate information/knowledge between neighborhoods, sites, cities, and regions. UrbanAI will facilitate urban adaptation to climate change at neighborhood, city, and regional scales. Under the subsequent project phases, we will aim to expand the input data streams (satellite, drones, and crowdsourcing using specifically developed mobile apps). Phase I: Our project will: Design API interfaces and SQL databases for collecting and managing urban data. Establish workflows for integration and management of data and information generated under the Urban Integrated Field Laboratories (UIFLs) projects funded by DOE-BER. Develop a computationally inexpensive thermal model to simulate buildingsĺ energy requirements. Develop, test, and validate ML techniques for characterization, analyses, and prediction of urban data and model simulations. Develop a community engagement plan (CEP). Design the UrbanAI software in a way that will provide: processing of real-time data streams, data storage and curation, data analytics and visualization, AI/ML interpretations and predictions, and dissemination of the obtained results to the users using web and IoT apps. UrbanAI will be explicitly designed considering customers in the utility, energy, manufacturing, and technology sectors. The project's technical, economic, and social benefits are nearly unlimited and far-reaching. Our product will enable customers to more efficiently and effectively visualize and analyze urban datasets from existing in situ networks, remote sensing (ground-based, airborne, drone, satellite), models (including digital twins), crowdsourcing, and other non-traditional data sources (such as IoT and edge devices) to make them easier to access by research and user communities.